Rosa Arboretti, Nicolò Biasetton, Riccardo Ceccato, Alberto Molena, Enrico Peruzzo, Luigi Salmaso, Alessio Sani

Abstract:

Football is one of the most widely followed sports in Italy, and football clubs have now become fully-fledged businesses. In this context, it is of utmost importance to be able to predict match attendance. In this work, we propose a two-phase framework. In the first phase, through a survey containing a conjoint analysis, we will be able to identify fan preferences related to certain match attributes, such as weather, start time, and the day of the week on which the game is played. From the responses we gather, we will create a variable called “utility”, which is then included in the second phase, represented by the development of ML models with the aim of predicting stadium attendance. We have evaluated this approach in collaboration with an important Italian football Team playing in the first division, using historical data for the training and testing phase. Moreover, we have also applied the proposed methodology to a match of the current season (2023-24). Generally, we can say that the proposed framework performs quite well, allowing us to predict stadium attendance in advance.

Keywords: Football, Conjoint analysis, Machine Learning.

Date Published: February 27, 2024
DOI: 10.11159/jmids.2024.004

View Article